Yves BERTRAND, EDF R&D (France),
[email protected] Delphine
LAURICHESSE, EDF R&D (France),
[email protected]
ABSTRACT
that EDF has developed is well adapted for replacing
mineral oils in medium voltage equipment. The comparison
of the evolution of its main properties with those of
standard
mineral oil, during ageing in contact with cellulosic
materials
and metal representative of those used in transformer,
shows that the behaviour of this new fluid is very close to
that of standard mineral oil.
KEYWORDS
INTRODUCTION
In collaboration with transformer manufacturers and an oleo-
chemistry research centre, EDF has developed a new dielectric fluid
based on modified vegetable oil (natural esters) which can be used
as an insulating and cooling liquid in distribution transformers
[1]. The formulation of this fluid has been optimised in order to
improve physical, chemical and electrical properties of natural
esters, and then to achieve characteristics close to those of
mineral insulating oils. In particular, this liquid shows
viscosities hardly higher than mineral oil ones, and pour point
close to -30°C. However, because of its fire point around 200°C, it
may not be classified among non-flammable liquids, as pure
vegetable oils are. On the contrary, the environmental features of
this vegetable based liquid stay equivalent to those of natural
esters, i .e. far better than mineral oil ones.
The oxidation stability assessment of natural dielectric liquids
needs to be compared to the mineral oil performance. But the
standard test methods, dedicated to either petroleum products or
fats, do not allow relevant judgment on the future behaviour of
vegetable esters inside real distribution transformers. So,
additional laboratory testing has been carried out for a better
understanding of the degradation consequences in transformers
filled with this new dielectric liquid. These tests consisted in a
thermal ageing of oils in contact with various materials
representative of those existing in the transformers.
EXPERIMENTAL
Investigations have been performed in parallel with both a standard
insulating mineral oil (paraffinic type, uninhibited) and the EDF
vegetable composition. This vegetable based composition is made by
blending an oleic rapeseed oil (i.e. triacylglycerols =
glycerol esterified
by 3 fatty acids) with esters prepared from an mono-alcohol and the
fatty acids of the same rapeseed oil (by transeste-
rification).
In the main test setup, we have used pressboards (thickness 0.5 and
1 mm) and wrapping paper ribbon (width 2 cm), i.e. cellulosic
materials, all added in oil in proportions (w/w) close to those in
typical distribution transformers. An unvarnished cooper wire
(diameter 1 mm) was used as oxidation catalyst during ageing (55
mm²/goil). In some testing combinations, epoxy printed paper, paper
with polyester film, bakelite and metallic specimens (aluminium,
enamelled copper and magnetic steel) have been also placed in the
test vessels.
Oils and other material samples were dried (16 hours under vacuum
at 80°C) separately before arrangement in the test vessel. Another
drying step is carried out after oil impregnation of the paper (6
hours under vacuum at 80°C), immediately followed by the ageing
period. The ageing took place in a lab oven at 120°C during 7 days
and under dry air (without bubbling in the oil), except for one
vessel kept at room temperature during 7 days.
Test Ref.
Oil Cu
catalyst Papers
Other materials
PA Veget.
@ 20°C
VA Veget.
1 Veget.
3 Veget.
5 Veget.
7 days @ 120°C
Table 1 : Oil/paper/other material combinations tested
After ageing, the different measurements performed are: o on
the oil : water content, acidity, furan derivative content,
breakdown voltage and FDD (tan δ) ; o on the papers : water
content, DPv (viscosimetric
polymerisation degree) and tensile strength.
Cellulosic materials
Water content
The moisture in the different papers after drying was lower than
0.5 %, representing typical water content of cellulosic materials
in new transformers. In the wrapping paper and 0.5 mm pressboard,
the moisture was 0.35 % (see Table 2). After 7 days in
ambient room atmosphere, pressboard water contents reach 2.2 % in
vegetable oil (180 ppm in oil), and 1.7 % in mineral oil (9 ppm in
oil) – cf. tests A and B in Table 1. In these cases the water
comes mainly from the atmosphere, and the data illustrate the
higher water solubility of vegetable oil.
Test Ref. NA NB 3 4 7 8
Oil Veget. Mine. Veget. Mine. Veget. Mine.
Metal - Cu All materials
Oil 75 < 5 98 12 86 17
Table 2 : Water contents in 0.5mm pressboard (%) and
oils (ppm), before [N columns] and after ageing
After thermal ageing, under dry air, Table 2 shows that
moisture contents in paper stay under the values reach when test
vessels are left at the free atmosphere, and also that they are
lower in vegetable oil than in mineral one.
Polymerisation degree
The unused 0.5 mm pressboard and wrapping paper have initial DPv
equal to 1 270 and 1 090, respectively.
Test Ref. PA PB 3 4 7 8
Oil Veget. Mine. Veget. Mine. Veget. Mine.
Metal - Cu All materials
1 260 1 220 930 870 1 030 1 020
wrapping paper
Table 3 : DPv of 2 papers after ageing
The DPv data of Table 3 confirm the influence of the copper and
other metals on the cellulose depolymerisation, but do not show
very significant differences between the results in EDF vegetable
composition and the mineral oil. Never- theless, the thin wrapping
paper appears to be more sensitive to the ageing than the
pressboard (DPv reduction about 40% against 29% approximately in
cases 3 and 4). And, it can be seen a lower degradation when
aluminium and magnetic steel are present with copper.
Tensile strength
The tensile strength was measured with a mechanical test setting
characterised by jaws placed edge to edge when clamping the paper
sample. This test is in common use in paper manufacturing. It gives
results more correlated to the state of the cellulose fibres, than
the electrotechnical standard tests which concern the structure of
the paper (corresponding to the arrangement of the fibres).
The results obtained (not shown) are consistent with the DPv
measurements. They show no differences between samples aged in
vegetable and mineral oils. However, we measure a loss of only few
% on the tensile strength, indicating that the cellulose fibres are
still strong in spite of the significant decrease of the DPv.
Furan derivatives in oils
It is known that the degradation of cellulosic materials leads to
the formation of these compounds. They are partially soluble in oil
where they can be measured by HPLC method. The results are
presented in the Table 4. After ageing 7 days at 120°C with
metal catalysts, furan derivatives can be detected in both oils,
and the outstanding feature is the much higher amount of
furfurylalcohol produced in the vegetable oil. The quantities of
the other derivatives are comparable, even lower in vegetable than
in mineral oil. The formation of this specific compound is not
explained but probably related to the chemical composition of the
two oils.
Test Ref. 3 4 7 8
Oil Veget. Mine. Veget. Mine.
Metal Cu All materials
5HMF : 5-hydroxymethl-2-furfural ; 2FAL : 2-furfural ; 2ACF :
acetylfuran ; 5MEF : 5-methyl-2-furfural ;
2FOL : 2-furfurylalcohol.
Oils
Water content
The water contents of vegetable and mineral oils was, respectively,
120ppm and 10 ppm before the drying step, and 75 ppm and under 5
ppm after. The Table 2 shows that the moisture in oil does not
increase very much when papers are present in the ageing test. The
water contents of oils increase much more when no paper is present
during ageing as shown in Table 5 below.
Test Ref. 1 2 5 6
Oil Veget. Mine. Veget. Mine.
Metal Cu All materials
Table 5 : Water content (ppm) in oils after ageing
without papers in test vessels
Acidity
Oil Veget. Mine. Veget. Mine. Veget. Mine.
Metal - Cu All materials
before [N columns] and after ageing
Typically in our tests, the acidity of mineral oil is more than one
order of magnitude higher after ageing, while it is multiplied by
less than 2 in vegetable composition.
Breakdown voltage and dielectric dissipation factor
Table 7 shows that the breakdown voltages of mineral oil are always
lower than the vegetable oil ones, and the dispersion of the
measurements (done in accordance with IEC 60156) is also much
higher in mineral oil than in vegetable one – 20 to 30 % against 5
to 10 %.
Test Ref. NA NB 5 6 7 8
Oil Veget. Mine. Veget. Mine. Veget. Mine.
Metal - All materials
DDF (tan δ)
Table 7 : Breakdown voltages (kV) and DDF of oils,
before [N columns] and after ageing
On one hand, the dielectric dissipation factor, DDF (tanδ), is
known to be higher in natural esters than in mineral oil. But on
the over hand, the data in Table 7 show that its evolution, after
ageing, is slower in the vegetable composition tested.
DISCUSSION
Concerning the moisture content of the test vessels, i.e. the
sum of the water in oil and in papers (when present), we can
calculate that it is higher after ageing in mineral oil than in
vegetable one (see Table 8). Moreover, without papers during the
ageing tests (cases 1- 2, 5-6 – cf. Table 1), the water
content in oil is multiplied by more than 8 in mineral oil, and by
approximately 3 in vegetable EDF composition.
Test Ref. NA NB 3 4 7 8
Oil Veget. Mine. Veget. Mine. Veget. Mine.
Metal - Cu All materials
Total 180.6 175.4 507.4 750.9 606.5 751.3
Table 8 : Water contents (mg) in oils and papers
Table 8 shows that the water content of the paper is systematically
lower in vegetable oil than in mineral oil. This feature is related
to the much higher solubility of moisture in esters, and it has
been shown that this will increase the useful life of the
cellulosic insulation in transformers [2].
CONCLUSION
The vegetable ester composition that EDF has developed presents
functional characteristics that match the technical specification
for insulating and cooling fluid in low and medium voltage
transformers. Moreover, our investigations show that, the evolution
during ageing of the EDF composition characteristics do not differ
from standard mineral oil ones. They differ in the baseline values
and magnitude, but not in the trends of the evolutions. The acid
number and water content, higher in the vegetable oil, tend to
increase less quickly than they do in mineral oils. So the paper
ageing in the EDF fluid, which appears to be at least equivalent
that in mineral oil, should be slower in transformer taking into
account the results presented here. Compared to other natural ester
dielectric fluids (i.e. more or less pure triacylglycerol
matrix), the EDF vegetable esters composition is very close to
standard mineral oils by its physical and thermal properties, and
appears to behave in the same way in real transformer environment.
So, it is well suited to replace standard mineral oil in medium
voltage transformers, without any modifications of the design and
maintenance of these equipments. EDF is now performing
investigation in real scale on its distribution network. Several
sealed transformers, built by various manufacturers and whose
powers lie between 50 and 400 kVa, are currently tested after
integral filling with EDF vegetable dielectric composition.
ACKNOWLEDGMENTS
The authors thank VALAGRO (Mr L.C. Hoang) for providing the
vegetable composition used in this study, and the ADEME who
supports this work. The authors thank also their colleagues for
their review and suggestions.
REFERENCES
[1] Y. Bertrand and L.C. Hoang, 2004, "Vegetable oils as
substitute for mineral oils in medium voltage equip-
ment ", Proceedings CIGRÉ 2004, D1-201. [2] C.P. McShane et
al ., 2006, "Natural ester dielectric fluid
development ", Proceeding IEEE 2005-2006 PES T&D.